Solving the small sample size problem in face recognition using generalized discriminant analysis

The goal of face recognition is to distinguish persons via their facial images. Each person's images form a cluster, and a new image is recognized by assigning it to the correct cluster. Since the images are very high-dimensional, it is necessary to reduce their dimension. Linear discriminant analysis (LDA) has been shown to be effective at dimension reduction while preserving the cluster structure of the data. It is classically defined as an optimization problem involving covariance matrices that represent the scatter within and between clusters. The requirement that one of these matrices be nonsingular restricts its application to datasets in which the dimension of the data does not exceed the sample size. For face recognition, however, the dimension typically exceeds the number of images in the database, resulting in what is referred to as the small sample size problem. Recently, the applicability of LDA has been extended by using the generalized singular value decomposition (GSVD) to circumvent the nonsingularity requirement, thus making LDA directly applicable to face recognition data. Our experiments confirm that LDA/GSVD solves the small sample size problem very effectively as compared with other current methods.     

应用金属迭片挠性联轴器改造D1000-32型风机

使用液力偶合器因故障频繁，对高炉稳产、高产造成严重威胁，而且维修费用较高，对偶合器改造后，应用金属迭片挠性联轴器其故障率、维修费用都很低。     

High training set size reduction by space partitioning and prototype abstraction

Instance-based learning methods like the nearest neighbour classifier generally suffer from the indiscriminate storage of all training instances, resulting in large memory requirements and slow execution speed. In this paper, new training set size reduction methods based on prototype generation and space partitioning are proposed. Experimental results show that the new algorithms achieve a very high reduction rate with still an important classification accuracy.     

钛精矿回转窑直接还原过程结圈可能性的探讨

通过煤种，入窑球冷态强度和还原过程强度，以及钛精矿与煤灰混合物的软化温度等分析探讨了钛精矿预热球团回转窑直接还原产生结圈的可能性，获得了可以避免结圈的结论。     

The Difference in Turbulent Diffusion between the Active and Passive Scalars in a Thermally Stably Stratified Medium

The difference in the turbulent diffusion between the active (heat) and passive (mass) scalars in a thermally stably stratified medium is investigated. The axisymmetric problem is treated on the formation of a turbulent circulation flow above a heated disk and on the turbulent diffusion of a passive scalar (impurity) from a continuous surface source in a stably stratified medium. The results indicate that the thermal stratification causes appreciable differences in the coefficients of turbulent transfer between the active (heat) and passive (mass) scalars. This means that the assumption of the identity of the coefficient of turbulent diffusion of heat and mass, employed in conventional models of turbulence, produces significant errors in estimating the heat and mass transfer in a thermally stably stratified medium.     

A Theoretical Framework for Core Material Properties Identification in Cellular Solids using Novelty Detection

Abstract:  This paper presents a novelty detection-based technique to identify core material properties of honeycombs and cellular structures. A numerical model (finite element) representing full scale and/or reduced size of the cellular solid is used to generate transmissibilities between topological homologous points at cells in different locations. In order to make the procedure robust against noise, these transmissibilities are artificially corrupted. This step is representative of a series of experimentally obtained measurements which automatically return information relating to the size and shape of data noise along with the mean measurement. The next stage, both in this paper and in the proposal for the experimental approach, is to generate several further sets of transmissibilities from the finite element model; the only difference being that the core material properties are altered from the original values. A novelty detection framework is then adopted to find a similarity measure between each of these 'test' transmissibilities and the original set thereby identifying the material properties. Although this work is concerned with identifying only one material property, the methodology extends to identifying several properties.     

Product Evaluation Knowledge Acquisition Based on Rough Sets

1 Introduction Acquiring knowledge is constantly encountered in the mechanical product evaluation process, such as during setting up evaluation index system and determining evaluation knowledge. The most useful expression form of knowledge is production rule whenever concerning the field knowledge or decision knowledge. How to effectively acquire and express the knowledge becomes the issue that the evaluation work must solve at first. One of an important application of the rough set theory is …     

珠钢不锈钢冶炼技术探讨

本文阐述了不锈钢冶炼的关键技术，为生产准备提供帮助。     

化学还原法制备纳米铜粉的研究

本文采用KBH4在液相中化学还原CuSO4,并加入KOH和络合剂EDTA ,制得了纳米级的纯净的铜粉 ,通过调整反应物的浓度 ,可以消除Cu2 O等杂质。制备的纳米铜粉还存在一定程度的团聚 ,需试验加入分散剂来改善。     

Aero-Acoustics of Modern Transonic Fans—— Fan Noise Reduction from Its Sources

The noise of aerodynamics nature from modern transonic fan is examined from its sources with the perspective of noise reduction through aero-acoustics design using advanced Computational Fluid Dynamics (CFD) tools. In particular the problems associated with the forward propagating noise in the front is addressed. It is identified that the shock wave spillage from the leading edge near the fan tip is the main source of the tone noise. Two different approaches have been studied to reduce the forward arc tone noise and two state-of-art transonic fans are designed using the strategies developed. The following rig tests show that while the fans exhibit other noise problems, the primary goals of noise reduction have been achieved through both fans and the novel noise reduction concept vindicated.